JP2005057813A - Image scanner - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a new image scanner which can detect correctly an edge which image data read by a line image sensor have. <P>SOLUTION: Then image scanner comprises a binary-coded means deriving two binary-coded values per pixel by binary-coding with a distinct threshold while the image data are binary-coded with a certain threshold, an extraction means which extracts the binary-coded value in a pixel which belongs to a middle picture element block defined alternatively in form comprising only a noticed pixel or a noticed pixel and the binary-coded value in a pixel which belongs to two picture element blocks defined by form which adjoins the middle picture element block, and a determining means to determine whether a pixel position of the noticed pixel is an edge according to the binary-coded value extracted by the extraction means. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to an image reading apparatus capable of accurately detecting an edge portion of image data read by a line image sensor.

  In consideration of the demand for high-speed document reading, an image reading apparatus that reads a document that is being transported or placed on the document is designed so that the document can be read without performing complex image processing. A method of restricting the posture to a certain one is used.

  That is, when a plurality of read originals are set on the automatic paper feed mechanism and read continuously, in the conventional technique, the paper guide of the automatic paper feed mechanism is matched to the size of the read original. A method is used in which the position of the original is restricted to a predetermined position and is read by setting the position.

  Further, in the case where a read document is placed on a document table and read, in the related art, the user can place the read document in a specified posture on the document table, thereby changing the posture of the document. It uses a method of restricting and reading certain things.

  However, according to such a conventional technique, there is a problem that the original cannot be read accurately if the attitude of the original does not become a specified one.

  In other words, if the original is set to be tilted, the image data of the tilted original is input to the image reading apparatus, so that the original cannot be read accurately.

  Certainly, due to recent advances in image processing technology, it has been widely performed to perform editing processing such as rotation processing on image data using tilted image data as a processing target on a personal computer or the like.

  However, the rotation processing realized by such an image processing technique takes a long processing time, and therefore cannot be applied to an image reading apparatus that requires real-time document reading.

  For this reason, according to the conventional technology, for example, an A4 size read original and a B5 size read original are mixedly read using an image reading apparatus having an automatic paper feeding mechanism. Then, the orientation of the smaller B5 size read original is not tilted because it is not restricted by the automatic paper feed mechanism, and this makes it impossible to read the B5 size read original accurately. In addition, there is a problem that high-precision image reading cannot be realized because there is a problem that a B5-size read original becomes an A4-size image having a blank.

  In order to solve this problem, the line image sensor is used to read a transported document or a placed document, and an edge portion of image data read by the line image sensor is detected and detected. Based on the result, it is conceivable to use a method for performing processing for solving this problem.

  Conventionally, when this edge portion is detected, a threshold value is set in advance, and the binarized value is obtained by binarizing the pixel data of each pixel using the threshold value, and the binarization is performed. A portion where the value changes is detected as an edge portion.

  However, when the edge portion is detected according to such a simple method, there is a problem that the edge portion cannot be accurately detected due to the influence of the shadow of the original or noise.

  Therefore, it is necessary to establish a technique for accurately detecting the edge portion of the image data read by the line image sensor.

  The present invention has been made in view of such circumstances, and an object thereof is to provide a new image reading apparatus that can accurately detect an edge portion of image data read by a line image sensor.

  In order to achieve this object, the image reading apparatus of the present invention detects an edge portion of image data read by a line image sensor, binarizes the image data with a certain threshold value, and Binarization means that obtains two binarized values for each pixel by binarizing with different thresholds, and 2 of pixels belonging to the middle pixel block defined in a form including only the target pixel or including the target pixel Extracting means for extracting a binarized value and binarized values of pixels belonging to two pixel blocks defined adjacent to the middle pixel block, and a binarized value extracted by the extracting means And a determination unit that determines whether or not the pixel position is an edge portion.

  In the image reading apparatus of the present invention configured as described above, in the case where the background and background of the document have contrast, in order to detect the edge portion by detecting the boundary portion, The binarized values of all pixels belonging to one pixel block adjacent to the block indicate the same value, and the binarized values of all pixels belonging to the other pixel block adjacent to the middle pixel block are When the same value of a different value is indicated, processing is performed so as to determine that the pixel position of the target pixel is an edge portion.

  When the background and background of the document do not have contrast, the determination means detects the shadow generated at the boundary between the document and the background, so that the determination means 2 is adjacent to the middle pixel block. When the binarized values of all the pixels belonging to one pixel block indicate the same value and at least the binarized value of the target pixel belonging to the middle pixel block indicates the same value different from that Processing is performed so that the pixel position of the pixel is determined to be an edge portion.

  In this way, in the image reading apparatus of the present invention, in order to be able to detect an edge portion without being affected by shadows or noise, the concept of a pixel block is introduced and pixel data of each pixel is converted into two A binarized value is obtained by binarizing with a threshold value. When the binarized value is used and the background and background of the document have a contrast, the ambiguity of the boundary portion is taken into consideration. The edge portion is detected by judging whether one of the two pixel blocks adjacent to the boundary portion shows a clear background of the original and the other shows a clear background. To process.

  If the background and background of the document do not have contrast, considering the shade indicated by the boundary portion, both the two pixel blocks adjacent to the boundary portion are clearly defined. And processing to detect an edge portion by determining whether or not the pixel block located at the boundary portion shows a clear state different from that.

  According to this configuration, according to the image reading apparatus of the present invention, when the conveyed document or the placed document is read by the line image sensor, the edge portion of the image data read by the line image sensor is detected. The edge portion can be accurately detected without being affected by shadows or noise.

  Hereinafter, the present invention will be described in detail according to embodiments.

  FIG. 1 shows an embodiment of an original reading apparatus 1 having the present invention.

  As shown in this figure, a document reading apparatus 1 according to the present invention includes a line image sensor 10 that reads a document conveyed by an automatic paper feeding mechanism, and an A / D converter that performs A / D conversion on image data read by the line image sensor 10. A D / A converter 11, a memory control unit 12 that reads image data by performing prescribed image processing using image data output from the A / D converter 11, and an A / D The image data output from the D converter 11 is input, the edge detection unit 13 that detects the edge portion of the image data, and the edge information detected by the edge detection unit 13 is input by using, for example, firmware. An inclination / size detection unit 14 for detecting the inclination and size of the image and a scanner, for example, configured to be firmware read from the memory control unit 12 Instructs the management start, it receives the detection result of the tilt / size detecting section 14, and a scanner control unit 15 for instructing the start of the deskew process to the memory control unit 12.

  The memory control unit 12 stores a first image memory 20 that stores document image data read by the line image sensor 10 in the reading order, and a document image data stored in the first image memory 20. An inclination correction reading unit 21 that reads out the image while correcting the inclination, a smoothing processing unit 22 that smoothes a jagged portion of image data generated by reading by the inclination correction reading unit 21, and image data processed by the smoothing processing unit 22 A smoothing processing unit 23 for smoothing the image, a horizontal enlargement processing unit 24 for enlarging the image data processed by the smoothing processing unit 23 in the horizontal direction, and an image data processed by the smoothing processing unit 23 in the vertical direction. A vertical enlargement processing unit 25, a binarization unit 26 for binarizing image data enlarged in the horizontal / vertical direction, and a frame portion of the document A masking unit 27 for removing existing shadows and noises, and a second image memory 28 for storing the image data read from the first image memory 20 subjected to these processes and delivering it to a host device (not shown). With.

  In this way, the memory control unit 12 does not correct the inclination after binarizing the image data of the document, but adopts a configuration in which the inclination is corrected in the state of the multi-value data before binarization. After that, a configuration is adopted in which smoothing processing is performed. Thus, the smoothing process becomes effective.

  As the smoothing processing unit 23, the horizontal enlargement processing unit 24, the vertical enlargement processing unit 25, the binarization unit 26, and the masking unit 27, those used in the conventional image processing technology can be used.

  First, an outline of each component included in the document reading apparatus 1 of the present invention will be described.

  (1) The edge detection unit 13 is provided to realize the present invention. When the edge detection unit 13 is activated in response to the activation of the scanner reading process issued by the scanner control unit 15, as shown in FIG. By checking the image data read by the line image sensor 10 in order from the left edge of the scan start for each line, it changes sharply from “white to black” or “black to white” without being affected by noise or shadow. A pixel position to be detected is detected, and the first pixel position to be detected is “left edge”, and the last pixel position to be detected is “right edge”, and this is reported to the inclination / size detection unit 14.

  (2) Upon receiving the “left edge coordinates” / “right edge coordinates” reported for each line from the edge detection unit 13, the inclination / size detection unit 14, as shown in FIG. While detecting the coordinates of the two vertices, a process of detecting the coordinates of the vertex which is the lowest end of the document is performed.

  Then, the inclination / size detection unit 14 calculates the “width” and “horizontal inclination (θ)” of the document shown in the drawing at the stage where the coordinates of the two vertices on the upper end side can be detected, and further, The horizontal magnification correction value is calculated according to “1 / cos θ”, and when the document conveyance speed is a prescribed standard, the same value as the horizontal magnification correction value is calculated as the vertical magnification correction value, and the document conveyance When the speed is not the specified standard, the vertical magnification correction value is calculated according to the ratio value of “1 / cos θ” and the conveyance speed, and their “width” / “slope (expressed in tan θ) ] / “Magnification correction value” and “cutout start address” defined as a storage destination address (storage address of the first image memory 20) of pixel data associated with the top leftmost vertex of the document Processing to be reported to the unit 15 is performed.

  Then, when the inclination / size detection unit 14 detects that the reading to the lowest vertex of the detected original has been completed (a state in which the “length” of the original shown in the figure can be calculated), the reading of the original is completed. A process of reporting the fact to the scanner control unit 15 is performed.

  (3) The scanner control unit 15 instructs the memory control unit 12 to execute the scanner reading process, and as shown in FIG. 4, the image data (multi-value data) of the original read by the line image sensor 10 Are written into the first image memory 20 of the memory control unit 12 in the order of reading.

  When the scanner control unit 15 receives the report of “cutting start address” / “width” / “tilt” / “magnification correction value” from the tilt / size detection unit 14, the “image processing mode” / The “masking range” is set, and the received value and the set value are notified, and the memory controller 12 is instructed to execute the deskew process.

  When the tilt / size detection unit 14 fails to detect the tilt / size, the scanner control unit 15 instructs the memory control unit 12 to execute a deskew process with the tilt “0”.

  When the scanner control unit 15 receives a notification from the inclination / size detection unit 14 that the reading of the document has been completed, the scanner control unit 15 responds to the process and instructs the memory control unit 12 to end the deskew process. Do.

  (4) When the scanner control process 15 is activated from the scanner control unit 15, the memory control unit 12 has the image data of the original read by the line image sensor 10 in the memory control unit 12, as shown in FIG. The first image memory 20 is written in the reading order.

  When the deskew processing is activated from the scanner control unit 15, the memory control unit 12 follows the “cutting start address” / “width” / “tilt” / “magnification correction value” notified from the scanner control unit 15. As shown in FIG. 5, the image data of the document stored in the first image memory 20 is read while correcting the inclination, and the jagged portion of the image data generated by the reading is smoothed. Further, the “image processing mode” notified from the scanner control unit 15 using the smoothing processing unit 23, the horizontal enlargement processing unit 24, the vertical enlargement processing unit 25, the binarization unit 26, and the masking unit 27 is performed. / Enter processing to perform image processing according to “masking range”, and complete deskew processing from scanner control unit 15 in conjunction with the end of document reading When instructed, it performs a process of ending the process.

  When the memory control unit 12 is instructed by the scanner control unit 15 to execute the deskew process with the inclination “0”, the image data of the original stored in the first image memory 20 is read as it is, so The correction or smoothing process is not performed.

  Next, a detailed configuration of each component constituting the document reading apparatus 1 of the present invention will be described.

  First, a detailed configuration of the edge detection unit 13 will be described according to an embodiment of the edge detection unit 13 shown in FIG.

  As shown in FIG. 6, the edge detection unit 13 includes a dot counter 130, a left end edge coordinate detection unit 131, a right end edge coordinate detection unit 132, an edge coordinate report unit 133, and an edge filter unit 134. Yes.

  The dot counter 130 resets the counter value to “0” if the horizontal synchronizing signal (shown in FIG. 2) is ON, and (2) if the horizontal valid signal (shown in FIG. 2) is ON. The image clock is counted up. (3) If the horizontal valid signal is OFF, the number of dots of image data sequentially input from the left end is counted by operating so as to hold the counter value.

  The left end edge coordinate detection unit 131 sets (0000) to the first edge register when the horizontal synchronization signal is ON, and (2) sets the horizontal valid signal to ON and sets “0000” to the first edge register. Is set, and the edge filter unit 134 detects an edge, the counter value of the dot counter 130 is set in the register of the first edge, and (3) the horizontal valid signal is OFF or the register of the first edge is set to “ If a value other than 0000 "is set or the edge filter unit 134 does not detect an edge, the value of the register of the first edge is held, so that the dot of the pixel detected as the first edge portion for each line. Holds the counter value.

  The right edge coordinate detection unit 132 sets (FF) to the last edge register when the horizontal synchronization signal is ON, and (2) the horizontal valid signal is ON, and the edge filter unit 134 detects the edge. If detected, the counter value of the dot counter 130 is set in the register of the last edge. (3) If the horizontal valid signal is OFF or the edge filter unit 134 does not detect the edge, the value of the register of the last edge is held. Thus, the dot counter value of the pixel detected as the final edge portion is held for each line.

  The edge coordinate reporting unit 133 (1) sets the dot counter value held in the first edge register to the register holding the left edge edge coordinate and the last edge register described above if the horizontal synchronization signal is ON. The dot counter value to be held is set in the register that holds the right edge coordinate. (2) If the horizontal synchronization signal is OFF, the value of the register that holds the left edge coordinate is held and the register that holds the right edge coordinate is held. By holding the value, the coordinates of the left end edge and the coordinates of the right end edge are reported to the inclination / size detection unit 14.

  The edge filter unit 134 uses a middle block centered on the pixel of interest, a right block located on the right side of the middle block, and a left block located on the left side of the middle block. In this case, in order to detect the boundary between the document and the document pressing plate and the edge portion of the document while removing noise and shadow, (1) as shown in FIG. When the pixel data of all the pixels belonging to it is smaller than the Low slice and the pixel data of all the pixels belonging to the right block are larger than the High slice, the target pixel position is detected as an edge, and (2) The pixel data of all the pixels belonging to the left block whose pattern is reversed is larger than the High slice, and the pixel data of all the pixels belonging to the right block is smaller than the Low slice. The target pixel position is detected as an edge when. Here, “Low slice <High slice” is set.

  Then, in the case of “original document pressing plate is white”, in order to detect the shadow of the document edge and the edge portion of the document while removing noise and shadow, (1) in FIG. As shown, when the pixel data of all the pixels belonging to the left block and the right block are larger than the high slice and smaller than the low slice of the pixel data of all the pixels belonging to the middle block, the target pixel position is edged (2) The pixel data of all the pixels belonging to the left block and the right block whose pattern is reversed are smaller than the Low slice and the pixel data of all the pixels belonging to the middle block When the pixel is larger than the high slice, the target pixel position is detected as an edge. Here, “Low slice> High slice” is set.

  FIG. 8 illustrates an embodiment of the edge filter unit 134.

  As shown in this figure, the edge filter unit 134 compares the pixel data (8-bit multi-value data) input from the A / D converter 11 with the Low slice / High slice described above, and the comparator 200. The shift register 201 that shifts and holds the high slice comparison output that is output from the shift register 202 that shifts and holds the low slice comparison output that is output from the comparator 200, and the pixels belonging to the left block held by the shift register 201 The AND circuit 203 that calculates the logical product of the comparison output values, the AND circuit 204 that calculates the logical product of the comparison output values of the pixels belonging to the left block held by the shift register 202, and the middle block held by the shift register 201 AND circuit 2 for calculating the logical product of the comparison output values of the pixels 5 and an AND circuit 206 that calculates the logical product of the comparison output values of the pixels belonging to the middle block held by the shift register 202, and the logical product of the comparison output values of the pixels that belong to the right block held by the shift register 201. The AND pixel 207, the AND circuit 208 that calculates the logical product of the comparison output values of the pixels belonging to the right block held by the shift register 202, and the calculation results of the AND circuits 203 to 208 are input, and the target pixel position is an edge. An edge determination circuit 209 for determining whether or not.

  When the edge determination circuit 209 receives the data input according to this configuration and determines that “the presser plate of the document is black” according to the filter type instructed from the outside, the filter type 1 shown in FIG. To determine whether or not the target pixel position is an edge portion, and according to the filter type instructed from the outside, it is determined that “the original press plate is white” Performs a filter type 2 logical operation shown in FIG. 9 to determine whether or not the target pixel position is an edge portion.

  Although not shown in FIG. 8, a circuit for generating a low slice / high slice is prepared, and this slice generation circuit is adapted to an appropriate low slice / high slice according to the filter type designated from the outside. Is generated and supplied to the comparator 200.

  The edge filter unit 134 allows the size of the middle block to be selected from 1 pixel / 3 pixels / 5 pixels / 9 pixels in order to cope with the unique characteristics of the scanner device, and the left block. The right block size can be selected from 2 pixels / 4 pixels / 8 pixels / 16 pixels.

  For example, as shown in FIG. 10, the selection configuration is such that the size of the left and right blocks is 2 pixels / 4 pixels / 8 pixels / 16 pixels by selecting an AND circuit by the selector with 1 pixel as the size of the middle block. The edge filter circuit that enables selection from among the three, the size of the middle block as 3 pixels, and the AND circuit is selected by the selector, so that the size of the left and right blocks is 2 pixels / 4 pixels / 8 pixels / 16 pixels Select edge filter circuit and 5 pixels as the size of the middle block and select the AND circuit by the selector to select the size of the left and right blocks from 2 pixels / 4 pixels / 8 pixels / 16 pixels Edge filter circuit to be able to do and left and right by selecting AND circuit by selector with 9 pixels as middle block size Edge filter circuit that makes it possible to select a lock size from 2 pixels / 4 pixels / 8 pixels / 16 pixels, and an edge by selecting a specified one from the output values of these four edge filter circuits This is realized by a selector input to the determination circuit 209.

  According to this configuration, when the edge detection unit 13 is activated in response to the activation of the scanner reading process issued by the scanner control unit 15, the image read by the line image sensor 10 for each line as shown in FIG. By checking the data in order from the left edge of the scan start, the pixel position that changes sharply from “white to black” or “black to white” is detected first without being affected by noise or shadow. The pixel position to be detected is the “left edge”, the pixel position to be detected last is the “right edge”, and the process of reporting it to the inclination / size detection unit 14 is performed.

  Next, detailed processing executed by the tilt / size detection unit 14 will be described.

  The inclination / size detection unit 14 performs processing for detecting the inclination and size of the document conveyed in an inclined posture as shown in FIG. At this time, as shown in FIG. 11B, even if the four corners of the document are bent or marks or the like are written at the four corners of the document, the document has a function of accurately detecting the inclination and size of the document. ing.

  Specifically, as shown in FIG. 12, the inclination / size detection unit 14 has an expression “Y = A * X + B” indicating the coordinate position of the left side of the document and an expression “Y = A * X + C ”, an expression“ Y = − (1 / A) * X + D ”indicating the coordinate position of the upper side of the document, and an expression“ Y = − (1 / A) * X + E ”indicating the coordinate position of the lower side of the document. This detection function is realized by calculating the coordinates of the two vertices on the upper end side of the document and the coordinates of the vertex that is the lowest end of the document according to these four equations.

  FIGS. 13 to 15 show an example of a processing flow executed by the tilt / size detection unit 14. Next, according to this processing flow, processing executed by the inclination / size detection unit 14 that realizes this detection function will be described in detail.

  When the tilt / size detection unit 14 is started by entering the document reading process, first, as shown in the processing flow of FIGS. 13 to 15, first, in step 1, the X coordinate of the edge is detected from the edge detection unit 13. After receiving the report of the “left edge coordinate” / “right edge coordinate” as a value and receiving the report, the process proceeds to step 2 to receive the reported “left edge coordinate” / “right edge coordinate”.

  Subsequently, in step 3, the Y coordinate value of the received “left edge coordinate” / “right edge coordinate” is calculated by adding the number of read lines, and in step 4, the received “left edge coordinate” is received. "/" Right edge edge coordinates "are stored in correspondence with the number of read lines.

  Subsequently, in step 5, it is determined whether or not two vertices at the upper end of the document have been detected. If it is determined that they have not been detected, the process proceeds to step 6 to receive the received "right edge coordinate". The document width is calculated by calculating a difference value between the “left edge coordinate” and the “left edge coordinate”.

  Subsequently, in step 7, it is determined whether or not the calculated document width is wider than the previous one, and when it is determined that the document is widened, the process returns to step 1 as it is. When making a determination, the process proceeds to step 8 where it is determined whether or not the state continues for a specified number of lines. If the specified number of lines is not continued, the process returns to step 1 as it is.

On the other hand, when it is determined in step 8 that the document width is not widened for a predetermined number of lines, that is, when it is determined that the position of the document width W shown in FIG. The left edge coordinate (X1, Y1) that is the starting point of the number of continuous lines and the left edge coordinate (X2, Y2) that is the arrival point of the number of continuous lines indicate the coordinate position of the left side of the document. Formula Y = A * X + B
And the right edge coordinate (X3, Y3) that is the starting point of the number of continuous lines and the right edge coordinate (X4, Y4) that is the arrival point of the number of continuous lines Formula indicating position Y = A * X + C
Is calculated.

  Subsequently, in step 10, the inclinations of the top and bottom sides of the document are “corresponding to the fact that the top side, the left side, and the right side of the document are orthogonal to each other and the bottom side, the left side, and the right side of the document are orthogonal to each other. -(1 / A) ".

  Subsequently, at step 11, edge coordinates located on the upper side of the document are extracted from the stored edge coordinates. The processing for extracting the edge coordinates located on the upper side of the document is, for example, smaller than the document width W determined to be reached in step 8 from the stored edge coordinates (left edge coordinate / right edge coordinate). Edge coordinates having the document width are extracted, and among them, an expression “Y = A * X + B” indicating the coordinate position of the left side of the document or an expression “Y = A * X + C” indicating the coordinate position of the right side of the document. This is done by identifying things that do not ride on.

Subsequently, in step 12, the extracted edge coordinates (for example, (X5, Y5) shown in FIG. 12) are expressed by the equation Y =-(1 / A) * X + D indicating the coordinate position of the upper side of the document.
By substituting into, the value of D is calculated for each of the extracted edge coordinates, and the average value of the calculated values of D is calculated while excluding the values deviating from the others. Expression indicating the coordinate position of the upper side of the document Y = − (1 / A) * X + D
Is calculated.

  Subsequently, in step 13, an expression “Y = A * X + B” indicating the coordinate position of the left side of the document calculated in step 9 and an expression “Y = − (1 / A / A) indicating the calculated coordinate position of the upper side of the document. ) * X + D ”is calculated to obtain the coordinates of the upper left vertex of the document, and the equation“ Y = A * X + C ”indicating the coordinate position of the right side of the document calculated in step 9 is calculated. The coordinates of the upper right vertex of the document are obtained by calculating the position of the intersection with the expression “Y = − (1 / A) * X + D” indicating the coordinate position of the upper side of the document.

  Subsequently, in step 14, it is detected from the calculated coordinates of the upper left vertex of the original and the coordinates of the upper right vertex whether the original is inclined upward or upward. That is, when the coordinates of the upper left vertex are closer to the reading start side than the coordinates of the upper right vertex, it is determined that the document is tilted in a left-up manner, and the coordinates of the upper right vertex are the coordinates of the upper left vertex. When the position is closer to the reading start side than the original, it is determined that the document is inclined in a right-up manner.

  Subsequently, in step 15, “the extraction start address (the storage address on the first image memory 20 of the pixel data associated with the edge coordinates of the upper left edge of the document)” defined by the one shown in FIG. , “Width”, “tilt (tan θ)”, and “magnification correction value (1 / cos θ)” are calculated and notified to the scanner controller 15, and the process returns to step 1.

  On the other hand, if it is determined in step 5 that the two top vertices of the document have been detected, the process proceeds to step 16 to determine whether or not the bottommost vertex of the document has been detected. When it is determined that the bottom vertex is not detected, the process proceeds to step 17 to calculate the document width by calculating the difference value between the received “right end edge coordinates” and “left end edge coordinates”.

  Subsequently, in step 18, when it is determined whether or not the calculated document width is approximately 0, and it is determined that the document width is not approximately 0, the process returns to step 1, and the document width is approximately When it is determined that the value is 0, the process proceeds to step 18 to extract the edge coordinates located on the lower side of the document from the stored edge coordinates.

  The processing for extracting the edge coordinates located on the lower side of the original is performed, for example, by the original width W after the detection of the original width W from the stored edge coordinates (left end edge coordinates / right end edge coordinates). Also, the edge coordinates having a small document width are extracted, and among them, the expression “Y = A * X + B” indicating the coordinate position of the left side of the document or the expression “Y” indicating the coordinate position of the right side of the document “ This is done by specifying something that does not ride "Y = A * X + C".

Subsequently, in step 19, the extracted edge coordinates (for example, (X6, Y6) shown in FIG. 12) are converted into an equation Y =-(1 / A) * X + E indicating the coordinate position of the lower side of the document.
By substituting into, the value of E is calculated for each of the extracted edge coordinates, and the average value of the calculated values of E is calculated while excluding values that are out of the others. Expression indicating the coordinate position of the lower side of the document Y = − (1 / A) * X + E
Is calculated.

  Subsequently, in step 20, the equation “Y = A * X + B” indicating the coordinate position of the left side of the document or the equation “Y = One of the formulas “A * X + C” is selected, and the intersection position of the selected formula and the formula “Y = − (1 / A) * X + E” indicating the coordinate position of the detected lower side of the original is determined. By calculating, the lowest vertex of the document is obtained.

  That is, if it is detected in step 14 that the document is tilted upward as shown in FIG. 12, the equation “Y = A * X + C” indicating the coordinate position of the right side of the document and the calculated document By calculating the position of the intersection with the expression “Y = − (1 / A) * X + E” indicating the coordinate position of the lower side of the document, the lowest vertex of the document is obtained, while the document is tilted upward to the right Is detected, the equation “Y = A * X + B” indicating the coordinate position of the left side of the document, and the equation “Y = − (1 / A) * X + E” indicating the calculated coordinate position of the lower side of the document, By calculating the position of the intersection, the lowermost vertex of the document is obtained.

  Subsequently, in step 21, it is determined whether or not the reading line coincides with the position of the lowermost vertex of the document. When it is determined that the reading line does not match, the process returns to step 1, and when it is determined that the reading line matches, Proceeding to step 22, the scanner control unit 15 is notified of the end of reading of the document, and the process ends.

  When it is determined in step 16 that the lowermost vertex of the document has been detected, the processing in steps 17 to 20 is omitted, and the process proceeds immediately to step 21 where the reading line is at the lowermost edge of the document. When it is determined whether or not the positions of the vertices coincide with each other, and when it is determined that they do not match, the process returns to step 1, and when it is determined that they match, the process proceeds to step 22 to end the reading of the document. 15 and the process is terminated.

  At the end of reading the original, the scanner control section 15 is notified of the “length” of the original shown in FIG. This is based on how much memory capacity (= [length / cos θ] * [width] for storing document image data for a host device (not shown) that receives document image data from the second image memory 28. This is to inform whether it is necessary to prepare / cos θ]).

  In this way, upon receiving the “left edge coordinates” / “right edge coordinates” reported for each line from the edge detection unit 13, the inclination / size detection unit 14 receives the upper edge of the document as shown in FIG. In addition to detecting the coordinates of the two vertices on the side, processing is performed to detect the coordinates of the vertex that is the lowermost end of the document.

  The inclination / size detection unit 14 detects the coordinates of the two vertices on the upper end side and the original “width” and the original “inclination (tan θ: preferably numerator 1, denominator is a natural number”. ”And the“ magnification correction value (1 / cos θ: when the document transport speed is not a standard one), the horizontal magnification correction value and the vertical magnification correction value are separately set. And “the extraction start address (the storage address on the first image memory 20 of the pixel data associated with the edge coordinates of the upper left edge of the document)” is specified. Processing to report to the scanner control unit 15 is performed.

  When the inclination / size detection unit 14 detects that the reading of the detected original to the lowest vertex is completed, the inclination / size detection unit 14 performs processing to report to the scanner control unit 15 that the reading of the original has been completed.

  As described above, the scanner control unit 15 instructs the memory control unit 12 to execute the scanner reading process, and as shown in FIG. 4, the image data (multivalued) of the document read by the line image sensor 10 is read. Data) is written into the first image memory 20 of the memory control unit 12 in the order of reading.

  When the scanner control unit 15 receives the report of “cutting start address” / “width” / “tilt” / “magnification correction value” from the tilt / size detection unit 14 according to the above-described processing, the “image processing mode” "/ Masking range" is set, and the received value and the set value are notified, and processing for instructing the memory control unit 12 to execute deskew processing is performed.

  When the scanner control unit 15 receives a notification from the inclination / size detection unit 14 that the reading of the document has been completed, the scanner control unit 15 performs a process for instructing the memory control unit 12 to end the deskew process.

  Next, details of processing executed by the memory control unit 12 will be described.

  When the scanner reading process is activated from the scanner control unit 15, the memory control unit 12, as shown in FIG. 4, stores the first image data of the document read by the line image sensor 10 in the memory control unit 12. A process of writing to the image memory 20 in the order of reading is performed.

  As will be described below, the memory control unit 12 sequentially reads the image data written in the first image memory 20 and processes it in parallel, so that the document read by the line image sensor 10 can be processed. The image data can be overwritten in the first image memory 20 in an overwritten manner. Thus, the first image memory 20 does not need to have a capacity for storing image data for one original.

  In response to this writing, the inclination / size detection unit 14 enters the above-described processing, and in response to this, the scanner control unit 15 instructs the memory control unit 12 to “cut out start address” / While performing notification of “width” / “tilt (tan θ)” / “magnification correction value” / “image processing mode” / “masking range”, execution of deskew processing is instructed.

  Upon receiving this deskew processing execution instruction, the inclination correction reading unit 21 constituting the memory control unit 12 notifies the notified “tilt (tan θ)” from the notified “cutout start address”. While the storage addresses are sequentially set in accordance with the “width”, the image data of the original stored in the first image memory 20 is read while correcting the inclination.

  For example, when a slope of “tan θ = 1/5”, which rises right by 5 pixels in the horizontal direction and shifts by 1 pixel in the vertical direction, is notified, as shown in FIG. Starting from the above, when 4 pixels are read in the horizontal direction, reading the previous reading line and reading 4 pixels in the horizontal direction are repeated for the length of the notified “width”. The first line is read out while correcting its inclination, and subsequently, the same horizontal position as the “cutting start address” is set for the second line, the third line, and the fourth line. Reading in the same way as the starting point, and entering the fifth line, repeatedly reading up to the last line of the document starting from the same position starting from the position shifted by one pixel to the right from the “cutting start address” Have It is, is performed a process of reading the image data of the document while correcting its tilt.

  On the other hand, when an inclination of “tan θ = 1/5” rising to the left is notified, reading out the next reading line after reading out four pixels in the horizontal direction from the “cutting start address”. The first line of the document is read while correcting the inclination by repeating the notified “width” length to read out four pixels in the horizontal direction, and then the second line is read. For the 3rd line, the 3rd line, and the 4th line, the same horizontal position as the “cutting start address” is read out in the same manner, and when entering the 5th line, “cutting start” By repeating the process of reading out in a similar manner starting from a position shifted by one pixel to the left from the “address” to the last line of the document, the document image data is read out while correcting its inclination. Ah .

  The storage address of the pixel data stored in the first image memory 20 has regularity as shown in FIG. 17, and from this, the inclination correction reading unit 21 reads “cutting start address” / “inclination (tan θ)”. ”/“ Width ”notification, you can know how to change the storage address and read the image data of the document with corrected tilt. The image data of the document stored in the first image memory 20 is read out while correcting the inclination while sequentially setting the storage addresses.

  FIG. 18 to FIG. 20 illustrate the image data reading process executed by the inclination correction reading unit 21 in the form of a processing flow.

  Here, in this processing flow, it is assumed that “horizontal inclination (1 / (p + 1))” and “vertical inclination (1 / (q + 1))” are given as different values. The process of determining whether or not it is going up to the right in step 2 is performed with a code added to “horizontal inclination (1 / (p + 1))”. Steps 3 to 14 are processes executed when the document is tilted upward to the right, and steps 15 to 26 are processes performed when the document is tilted upward to the left.

  If described with reference to FIG. 16, “p = 4” in FIG. The “address one line before the address corresponding to the p-th address following the address X” described in step 5 is, for example, the fourth pixel data e following the address X. This is the storage address of the pixel data f one line before that is associated with the storage address. Further, the “address after one line associated with the address next to the address X” described in step 14 is, for example, pixel data (1) (1 shown in the figure). Pixel data after one line (3) (in the figure) associated with the storage address of the next pixel data (2) (pixel data in which 2 is circled in the figure) (2) 3 is a storage address of pixel data surrounded by a circle.

  In the processing executed in steps 3 to 14 when the document is tilted upward to the right, the image data is read by tracing back one line before one line starting from the top left edge of the document. Thus, in the processing executed when the original is tilted to the left in steps 15 to 26, the image data is read by tracing one line and one line starting from the top left edge of the original. There is only a difference.

  When such an oblique reading method is used, an image shift locally occurs due to a quantization error of less than one pixel. A smoothing processing unit 22 constituting the memory control unit 12 is prepared for correcting this image shift.

  In order to notify the smoothing processing unit 22 to which pixel the smoothing processing should be performed, the inclination correction reading unit 21 shifts the pixel data in a form shifted in the vertical direction as shown in FIG. When reading, pixel data read immediately before the shift (for example, d in the figure) and pixel data read by the shift (for example, f in the figure) are defined as “horizontal misalignment pixels”, and the pixel data Processing is performed to turn on the horizontal shift bit of the 1-bit configuration to be added.

  Then, in order to inform the smoothing processing unit 22 to which pixel the smoothing processing should be performed, the inclination correction reading unit 21 shifts the pixels in the horizontal direction as shown in FIG. When reading data, the pixel data on the line read immediately before the shift (indicated by the left-upward slanted line in the figure) and the pixel data on the line read out of the shift (in the right-upward slanted line in the figure) Are defined as “vertical misalignment pixels”, and processing is performed to turn on the 1-bit vertical misalignment bits added to the pixel data.

  When the smoothing processing unit 22 constituting the memory control unit 12 receives the image data of the original whose inclination is corrected and read by the inclination correction reading unit 21, it is generated by the inclination correction using an image filter as shown in FIG. To make the jagged portion of the image data to be smoothed.

  That is, for a horizontal shift pixel in which the horizontal shift bit is turned on, for example, an image generated by tilt correction using an image filter (smoothing degree K can be selected) as shown in FIG. Performs processing to smooth the jagged portion of the data.

  Incidentally, in the image filter shown in FIG. 21A, when “K = 1”, the pixel data (X in the drawing) of the horizontal shift pixel is replaced with the average value of the pixel data of the pixels adjacent to the left and right. As a result, the process of smoothing the jagged portion generated by the horizontal shift pixel is performed.

  In addition, for a vertical shift pixel in which the vertical shift bit is turned ON, for example, an image generated by tilt correction using an image filter (smoothing degree K can be selected) as shown in FIG. Performs processing to smooth the jagged portion of the data.

  Incidentally, in the image filter shown in FIG. 21B, in the case of “K = 1”, the pixel data (X in the drawing) of the vertically shifted pixels is replaced with the average value of the pixel data of the vertically adjacent pixels. Thus, a process for smoothing the jagged portion generated by the vertical shift pixels is performed.

  In addition, for a horizontal / vertical shift pixel in which the horizontal shift bit is turned on and the vertical shift bit is turned on, for example, an image filter as shown in FIG. 21C (smoothing degree K can be selected) is used. Then, the process of smoothing the jagged portion of the image data generated by the inclination correction is performed.

  Incidentally, in the image filter shown in FIG. 21C, in the case of “K = 1”, the pixel data of the horizontal and vertical misalignment pixels (X in the figure) is the average value of the pixel data of the pixels adjacent vertically and horizontally. By replacing with, a process of smoothing the jagged portion generated by the horizontal and vertical misalignment pixels is performed.

  In order to enable application of this image filter, the smoothing processing unit 22 has a mechanism for holding image data for three lines read by the inclination correction reading unit 21.

  When the smoothing processing unit 23 included in the memory control unit 12 receives the image data of the document that has been subjected to the smoothing processing by the smoothing processing unit 22, the smoothing processing unit 23 uses the image filter as shown in FIG. A process is performed to smooth the jagged portions of the image data that did not exist.

  That is, for all pixels, for example, using an image filter as shown in FIG. 22 (the degree of smoothing K can be selected), image data that could not be removed by the smoothing processing by the smoothing processing unit 22 is used. Performs a smoothing process on the jagged parts of the image.

  Incidentally, in the image filter shown in FIG. 22, in the case of “K = 1”, the pixel data (X in the drawing) of the pixel to be processed is the average value of the pixel data of eight pixels adjacent in the eight directions. By replacing with, a jagged portion of image data that could not be removed by the smoothing process is smoothed.

  In order to enable application of this image filter, the smoothing processing unit 23 has a mechanism for holding image data for three lines subjected to the smoothing processing by the smoothing processing unit 22.

  The horizontal enlargement processing unit 24 constituting the memory control unit 12 reduces the image data of the original whose inclination is corrected read out by the inclination correction reading unit 21 by “cos θ” times in the horizontal direction compared to the original image data. Therefore, a process of enlarging the read image data by “1 / cos θ” times in the horizontal direction is performed.

  Then, the vertical enlargement processing unit 25 constituting the memory control unit 12 has the image data of the document whose inclination is corrected read by the inclination correction reading unit 21 multiplied by, for example, “cos θ” in the vertical direction compared to the original image data. Considering that the image data has been reduced (changes depending on the document conveyance speed), the read image data is enlarged by, for example, “1 / cos θ” times in the vertical direction.

  The horizontal enlargement processing by the horizontal enlargement processing unit 24 determines that it is necessary to insert one pixel for every 100 pixels in the horizontal direction when “1 / cos θ = 1.01”, for example. The average value of the pixel data of the left and right pixels at the insertion location is calculated, and the pixel data having the average value is inserted into the insertion location.

  The vertical enlargement processing by the vertical enlargement processing unit 25 determines that it is necessary to insert one line for every 100 lines in the vertical direction when, for example, “1 / cos θ = 1.01”. For example, the calculation is performed by calculating the average value of the pixel data of the pixels on the upper and lower lines of the insertion location, and inserting the pixel data having the average value into the insertion location.

  The binarization unit 26 constituting the memory control unit 12 compares the image data enlarged by the horizontal enlargement processing unit 24 and the vertical enlargement processing unit 25 with a prescribed threshold value, thereby converting the multivalued image data into 2 Performs the process of converting to a value.

  In this way, the memory control unit 12 does not correct the inclination after binarizing the image data of the document, but adopts a configuration in which the inclination is corrected in the state of the multi-value data before binarization. After that, a configuration is adopted in which smoothing processing is performed. Thus, the smoothing process becomes effective.

  The masking unit 27 constituting the memory control unit 12 performs a process of erasing shadows existing in the frame portion (edge portion) of the image data of the document binarized by the binarization unit 26, thereby performing the second processing. Processing to be stored in the image memory 28 is performed.

  In this way, when the scanner reading process is activated from the scanner control unit 15, the memory control unit 12 converts the image data of the original read by the line image sensor 10 into the memory control unit 12 as shown in FIG. Is written in the first image memory 20 in the reading order.

  When the deskew processing is activated from the scanner control unit 15, the memory control unit 12 follows the “cutting start address” / “width” / “tilt” / “magnification correction value” notified from the scanner control unit 15. As shown in FIG. 5, the original image data stored in the first image memory 20 is read while correcting its inclination, and the jagged portion of the image data generated by the inclination correction is made smooth. Further, smoothing processing is performed, and the “image processing mode” notified from the scanner control unit 15 using the smoothing processing unit 23, horizontal enlargement processing unit 24, vertical enlargement processing unit 25, binarization unit 26, and masking unit 27. ”/“ Masking range ”is entered, and the deskew processing is finished from the scanner control unit 15 in conjunction with the end of reading of the document. When instructed, it performs a process of ending the process.

  Although not described in FIG. 1, the memory control unit 12 includes an image processing unit, a processing unit that performs enhancement processing, a processing unit that performs error diffusion, and a processing unit that performs conversion to a floating slice. Etc. are prepared.

  As described above, the image reading apparatus of the present invention introduces the concept of a pixel block in order to be able to detect an edge portion without being affected by shadows or noise, and 2 pixel data of each pixel. Binarization is obtained by binarization with two threshold values, and when the background of the document has contrast with the binarization value, the ambiguity of the boundary portion is taken into consideration. The edge portion is detected by determining whether one of the two pixel blocks adjacent to the boundary portion shows a clear background of the original and the other shows a clear background. Is processed.

  If the background and background of the document do not have contrast, considering the shade indicated by the boundary portion, both the two pixel blocks adjacent to the boundary portion are clearly defined. And processing to detect the edge portion by determining whether or not the pixel block located at the boundary portion shows a clear state different from that.

  Thus, according to the image reading apparatus of the present invention, the edge portion can be detected without being affected by shadows or noise.

1 is an example embodiment of the present invention. It is explanatory drawing of a process of an edge detection part. It is explanatory drawing of the process of an inclination / size detection part. It is explanatory drawing of the storage process of the image data to a 1st image memory. It is explanatory drawing of a process of a memory control part. It is an example of an embodiment of a circuit configuration of the edge detection unit. It is explanatory drawing of the binarization process of an edge filter part. It is an embodiment of an edge filter unit. It is explanatory drawing of the determination process of an edge determination circuit. It is an embodiment of an edge filter unit. It is explanatory drawing of the original document conveyed. It is explanatory drawing of the detection process of an inclination / size detection part. It is an example of the processing flow which an inclination / size detection part performs. It is an example of the processing flow which an inclination / size detection part performs. It is an example of the processing flow which an inclination / size detection part performs. It is explanatory drawing of the read-out process of an inclination correction read-out part. It is explanatory drawing of the image data stored in a 1st image memory. It is explanatory drawing of the process which an inclination correction read-out part performs. It is explanatory drawing of the process which an inclination correction read-out part performs. It is explanatory drawing of the process which an inclination correction read-out part performs. It is explanatory drawing of the image filter which performs a smoothing process. It is explanatory drawing of the image filter which performs a smoothing process.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Document reader 10 Line image sensor 11 A / D converter 12 Memory control part 13 Edge detection part 14 Inclination / size detection part 15 Scanner control part 20 First image memory 21 Inclination correction reading part 22 Smoothing process part 23 Smoothing Processing unit 24 Horizontal enlargement processing unit 25 Vertical enlargement processing unit 26 Binarization unit 27 Masking unit 28 Second image memory

Claims (4)

In an image reading apparatus that detects an edge portion of image data read by a line image sensor,
Binarization means for obtaining two binarized values for each pixel by binarizing the image data with a certain threshold value and binarizing with a different threshold value;
The binarized value of a pixel belonging to a middle pixel block defined so as to include only the pixel of interest or including the pixel of interest and the pixel belonging to two pixel blocks defined in a form adjacent to the middle pixel block Extraction means for extracting the binarized value;
Determining means for determining whether or not the pixel position of the pixel of interest is an edge portion in accordance with the binarized value extracted by the extracting means;
A characteristic image reading apparatus. The image reading apparatus according to claim 1,
The determination means has the same binarized value of all pixels belonging to one pixel block adjacent to the middle pixel block, and belongs to the other pixel block adjacent to the middle pixel block. When the binarized values of all the pixels indicate the same value different from that, it is determined that the pixel position of the target pixel is an edge part.
A characteristic image reading apparatus. The image reading apparatus according to claim 1,
The determination means indicates that the binarized values of all pixels belonging to two pixel blocks adjacent to the middle pixel block have the same value, and the binary value of at least the target pixel belonging to the middle pixel block. Determining that the pixel position of the pixel of interest is an edge portion when the conversion value indicates the same value different from that,
A characteristic image reading apparatus. The image reading apparatus according to any one of claims 1 to 3,
The extraction means extracts the binarized value of a pixel belonging to a pixel block having a designated block size.
A characteristic image reading apparatus.

Images